Influenza viruses cause a highly contagious acute respiratory disease that has been responsible for epidemic and pandemic disease in humans for centuries. In the 20th century, influenza claimed millions of human lives in three different pandemics, i.e., 40 million worldwide deaths during the Spanish flu by the influenza strain H1N1 in 1918, 70,000 American deaths in 1957 by the influenza H2N2 strain, and 34,000 American deaths in 1968 by the influenza H3N2 strain. Currently, 5-20% of the United States population contract influenza annually, and more than 200,000 people become hospitalized as a result of complications, with approximately 36,000 mortalities per year (Fiore et al., 2008, Hoyert et al., 2005, Podewils et al., 2005). Depending on the antigen differences in nucleoprotein (NP) and matrix protein (M), influenza viruses are divided into three major subtypes, A, B and C.
Influenza A viruses are the major causes of human flu epidemics. Influenza A is subject to regular antigenic changes, brought about either by genetic drift and inter-type gene re-assortment (genetic shift). In both situations, prior immunity to influenza does not necessarily prevent infection with the new type, leading to localized epidemics or, in the case of genetic shift, a global pandemic of influenza. Influenza A is also responsible for all Avian flu.
Vaccines are the mainstay of prophylaxis against influenza, but there are significant technical and safety issues attendant to vaccines, including difficulties in predicting which viral strains may emerge, preparing sufficient quantities of vaccine, poor immunogenicity in the elderly and very young, and difficulty in storage and administration.
Influenza anti-viral drugs are an important adjunct to vaccination; however, substantial drug resistance has developed in influenza strains to two of the four currently approved anti-viral drugs (Hayden and Hay, 1992). Furthermore, only two anti-viral drugs (rimantadine and oseltamivir) are approved for chemoprophylaxis of influenza virus infection (Govorkova et al., 2001). The evidence for viral resistance to anti-viral agents indicates that more than one drug is necessary to effectively combat influenza. As a consequence of these persistent difficulties, new therapeutic and vaccine approaches for influenza are earnestly sought. In particular, the concept of a “universal vaccine” has generated much interest.
The M gene of influenza A encodes two proteins in overlapping frames: M1, the capsid protein and M2, an ion channel protein. Both M1 and M2 are highly conserved, with M2 encoding a small ectodomain (M2e) (Palese, 2006), making it a potential target for antibody-based immunity. The ability of a monoclonal anti-M2e antibody to reduce viral replication (Zebedee and Lamb, 1988) suggested M2e as a potential vaccine target. Slepushkin et al. demonstrated protection following M2e vaccination using baculovirus-expressed protein, with serum antibody responses detected against both amino- and carboxy-terminal M2e peptides, and presumed to be responsible for the protection against lethal challenge with a matched (H2N2) influenza A virus (Slepushkin et al., 1995). In 1999, Neirynck et al. described a “universal vaccine” based upon the 23 a.a. ectodomain of M2 and demonstrated protection against H3N2 challenge virus with an M2e sequence identical to or differing by one amino acid from the vaccine constructs (Neirynck et al., 1999). Multiple antigenic peptide (MAP) vaccines have also been shown to be protective (Mozdzanowska et al., 2003).
While most human H1 and H3 influenza viruses share complete homology with the M2e consensus sequence (termed “conM2” herein), M2e-specific antibodies have not been shown to bind to M2e peptides with considerable sequence divergence. In a study of M2e-carrier conjugate vaccines, serum antibodies specific for conM2 or the M2e sequence of A/PR/8/34 (A/PR8, H1N1) did not cross-react with M2e peptides from H5 and H7 avian viruses having 3 or 4 mismatches out of 24 a.a. (Fan et al., 2004). In another study, immunization with plasmid containing the entire M gene from A/PR8 was shown to protect against matched (H1N1) challenge, however there was limited evidence for M2-specific immune responses (Okuda et al., 2001, Watabe et al., 2001). Importantly, while a recent study used matched M2e peptide-liposome vaccines of various subtypes (Ernst et al.), none of the previously published work has documented protection against challenge with influenza viruses across substantial M2e sequence differences from the immunizing antigen and specifically against potential pandemic H5N1 influenza challenge. In contrast, we have obtained evidence that M2-specific antibody responses can potentially be broadly cross-reactive and protect against divergent influenza virus challenge (Tompkins et al., 2007). While vaccines based on M2 may become very useful, it is also possible that human or humanized antibodies recognizing M2 may be effective as passive vaccines or therapeutics.
The sequences of six murine anti-M2e mAbs, generated either by consecutive pulmonary infection (Mozdzanowska, et al., 2003) or immunization with purified M2 (Zebedee and Lamb, 1988), have recently been published (Zhang et al., 2006). These all have very similar recognition properties, with the recognized epitope located between amino acids 4-16 of the external portion of M2 (Zhang, et al., 2006). Interestingly, these mAbs all use the same VH, DH and JH genes, with minor differences (less than 7%) between them, and only two different kappa light chains. One of these antibodies, 14C2, recognizes M2e when expressed on the cell membrane after infection (Zebedee and Lamb, 1988), and reduces viral plaque size (Zebedee and Lamb, 1988) and viral production levels (Hughey et al., 1995) in vitro. Both 14C2 and another antibody, M2-80, have also been shown to have significant protective effects in mice (Mozdzanowska et al., 1999, Treanor et al., 1990, Zharikova et al., 2005).
The demonstrated anti-viral activity of 14C2 suggests that it would be a good candidate for humanization (Carter et al., 1992, Chothia et al., 1985, Hwang et al., 2005, Kettleborough et al., 1991, Pedersen et al., 1994, Roguska et al., 1994, Routledge et al., 1991, Studnicka et al., 1994, Tsurushita et al., 2005, Vargas-Madrazo and Paz-Garcia, 2003), a term describing a series of techniques in which the sequence of a murine antibody is changed so that it more closely resembles a human antibody sequence. Conceptually, this involves taking the binding loops of the murine antibody and grafting them onto a human variable region framework in such a way that they are still able to recognize the antigen of interest. This often involves the retention of some critical murine framework amino acids required to maintain the correct orientation of the binding site loops, as well as subsequent mutation and selection to maintain affinity. Humanization has been widely used, and nine approved drugs are humanized antibodies. Although a number of different methods to carry out humanization have been developed (Carter, et al., 1992, Chothia, et al., 1985, Hwang, et al., 2005, Kettleborough, et al., 1991, Pedersen, et al., 1994, Roguska, et al., 1994, Routledge, et al., 1991, Studnicka, et al., 1994, Tsurushita, et al., 2005, Vargas-Madrazo and Paz-Garcia, 2003), none of them has been demonstrated to be significantly superior to any other. Here we describe the humanization of the M2e-specific murine mAb, 14C2, demonstrate specificity for the native M2 protein, and confirm anti-viral activity of the humanized single-chain minibodies.
Antibodies provide an appealing strategy for the prevention or treatment of viral infections. Their specificity, relatively long half-life and limited toxicity are just a few of the strengths of this therapeutic modality. Although polyclonal antibodies are FDA approved for eight pathogens (hepatitis B, CMV, botulism, RSV, rabies, tetanus, VZV and vaccinia) (Zeitlin et al., 2000), there is a clear preference for therapeutics that are better defined. There were initial hopes that rodent mAbs could be used in therapy, but their immunogenicity has led to efforts to create mAbs which are more human in their sequences. With the advent of modern molecular biology, three main classes of mAbs with lower immunogenicity have been developed. These include chimeric antibodies, in which murine V regions are fused to human constant regions (Morrison et al., 1984), humanized antibodies, in which murine antigen binding loops are grafted onto human variable region framework sequences (Jones et al., 1986), and fully human antibodies, the latter being made by phage display (reviewed in Winter et al., 1994), or by applying traditional hybridoma technology to mice transgenic for the human immunoglobulin loci (reviewed in Lonberg, 2005). As chimeric antibodies retain some residual immunogenicity, humanized and human antibodies are most frequently used and are equally represented in clinical trials and approved drugs (Reichert et al., 2005).
Recombinant antibodies offer many advantages for the treatment of diseases (Reichert, et al., 2005), including those caused by infectious agents (Reichert and Dewitz, 2006), and viruses in particular (Marasco and Sui, 2007). Compared to antibodies produced in animals, they have greater potency, defined activity, lack infectious agents, avoid the development of serum sickness caused by immune responses to non-human antibodies, and with a half-life of up to 4 weeks, provide long periods of protection with relatively infrequent dosage schedules. For all indications, eighteen mAbs have received regulatory approval and over 150 are now in clinical development (Reichert and Dewitz, 2006, Reichert, et al., 2005). One mAb against respiratory syncytial virus, providing significant reduction in morbidity, has been approved for treatment of high risk pediatric cases and mAbs against over twenty other infectious agents, including SARS, rabies, West Nile virus, HIV, Dengue, Ebola, Hepatitis A, B and C, anthrax, E. coli and Staphylococcus, are under development (Marasco and Sui, 2007, Reichert and Dewitz, 2006). In addition to their therapeutic value, antibodies also have potential as passive vaccines, which can translate into months of protection following prophylactic administration: long enough to cover a flu season, or the community duration of a pandemic (Bartlett, 2006). As the means of production of human mAbs are well known, the process of manufacturing, as well as the necessary toxicology and clinical safety testing requirements are well understood. This results in a rapid development timeline, once suitable candidates have been identified. This is especially true for antibodies recognizing infectious epitopes, rather than human proteins, in which inadvertent unexpected reactions may occur (Feldman et al., 2000).
Influenza provides a number of viral targets for antibody therapies (Beigel and Bray, 2008). Antibodies to the hemagglutinin can neutralize the virus and readily prevent infection, however these antibodies are subtype and in many cases strain or Glade specific and so have limited use as antibody therapies (Simmons et al., 2007), even though the efficacy of the annual vaccines is related to their ability to induce HA antibodies. Neuraminidase antibodies, while not neutralizing may also protect against infection (Gillim-Ross and Subbarao, 2007). Unexpectedly, antibodies against the influenza nucleoprotein (NP), which coats the viral RNA have also been shown to be protective in mice (Carragher et al., 2008), although previous studies suggest NP immunization protects via T and not B cell responses (Epstein, 2003, Ulmer et al., 1993). Finally, the M2 protein has been widely explored as a target for both vaccines and drug therapies. M2 is an appealing target as it is expressed to high levels on virus-infected cells, it is relatively conserved compared to other surface viral antigens, and antibody responses to M2 proteins have been demonstrated to protect against human and avian influenza virus infections (Fiers et al., 2004, Tompkins, et al., 2007, Tripp and Tompkins, 2008, Wang et al., 2008). M2e, the M2 ectodomain is conserved at least in part because it is generated as a spliced transcript and the first 9 amino acids are shared by M1 capsid and M2 pore proteins (Palese, 2006).